What Is PGP? The Concepts, and the Actual Artefacts You'll See

The name, and why three terms get used almost interchangeably

PGP — Pretty Good Privacy — was created by Phil Zimmermann and first released in 1991, named, according to Proton's own account, after a fictional grocery store in Garrison Keillor's radio stories.^Proton Zimmermann's original program went through a genuinely tangled ownership history: PGP Inc., then Network Associates in 1997, then PGP Corporation in 2002, then Symantec in 2010, then part of a Broadcom acquisition in 2019 — and the Pretty Good Privacy trademark itself was formally abandoned in April 2020.^{TechTarget, Apr 2025} Because of that history, "PGP" today mostly refers to the open standard the original program's message format became, rather than to any specific company's product. OpenPGP is that open standard, published through the Internet Engineering Task Force since 1996 and still under active development.^{TechTarget, Apr 2025} GnuPG (GPG) is the most widely used free, open-source program that implements that standard.^{TechTarget, Apr 2025} In practice, "PGP" is used loosely to mean all three — the original brand, the open standard, and the tool most people actually run — and context usually makes clear which one a given sentence means.

Why PGP uses two different kinds of encryption at once

Public-key (asymmetric) encryption solves a real problem — you can encrypt a message for someone using only their public key, with no prior shared secret needed — but it's computationally expensive for large amounts of data. Symmetric encryption is fast but requires both sides to already share a secret key, which is exactly the problem you don't have a secure channel to solve if you're communicating with a stranger for the first time. PGP's actual process resolves both limitations at once: it generates a random, single-use symmetric session key, encrypts your actual message with that fast symmetric key, and then encrypts only that short session key using the recipient's public key.^Proton The recipient reverses the process — their private key decrypts the small session key, and that session key decrypts the actual message — getting the speed of symmetric encryption for the bulk of the data and the no-prior-contact convenience of public-key encryption for the part that actually needs protecting in transit.^{TechTarget, Apr 2025}

What a PGP digital signature actually is

A signature is a separate function from encryption, and a message can be signed, encrypted, or both independently. To sign a message, PGP generates a hash — a fixed-length mathematical digest of the message content — and encrypts that hash with the sender's own private key.^{TechTarget, Apr 2025} Anyone holding the sender's public key can then decrypt that hash and independently compute their own hash of the message they received; if the two match, two things are proven at once: the message came from whoever holds the private key matching that public key, and the message wasn't altered after signing, since even a single changed character produces a completely different hash.^Proton What a valid signature does not prove is a separate question this page won't paper over: it proves authorship and integrity, never honesty — a signature confirms who wrote something, not that what they wrote is true, a distinction covered fully in practice on PGP on the dark web.

What these things actually look like

A public key, once exported in text form, looks like this in outline — the real thing runs many lines longer, but the shape is what matters here:

-----BEGIN PGP PUBLIC KEY BLOCK-----

mQINBGX1... [several lines of base64-encoded key data] ...AEBAAG0
=Ab3d
-----END PGP PUBLIC KEY BLOCK-----

A signed message — specifically a cleartext signature, which keeps the message readable without decryption while still carrying a verifiable signature block — looks like this in outline:

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

The message text goes here, exactly as written.
-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEE... [several lines of base64-encoded signature data]
=xY9k
-----END PGP SIGNATURE-----

And a successful verification, run from the command line, reports something close to this:

gpg: Signature made [date] using RSA key [key ID]
gpg: Good signature from "Name <email>" [full validity]

The word to look for is "Good." A bad signature, an expired key, or a key gpg doesn't recognize at all produces a visibly different, unmistakable warning instead — never a message that looks almost the same. PGP on the dark web shows what each of those failure cases actually looks like and what to do about each one.

Two terms are worth telling apart before you go further: a key ID is a short, abbreviated identifier for a key — often just the last 8 or 16 characters of the fingerprint — while the fingerprint is the full-length identifier, generally 40 hexadecimal characters for the keys covered on this site. A key ID is convenient for a human to type or read aloud; it is also short enough that a determined attacker could plausibly generate a colliding key sharing the same short ID, which is precisely why every verification instruction on this site insists on the full fingerprint rather than the shorter, more collision-prone ID.

The web of trust, and why it's not the only model

Ordinary web browsing verifies identity through certificate authorities — centralized organizations that vouch for who owns a domain, which the browser trusts by default. PGP's original model is decentralized instead: a user can personally verify someone else's key — checking a fingerprint in person, confirming identity documents — and then sign that person's key to say so publicly.^{TechTarget, Apr 2025} If you trust someone's judgment about vetting keys, you can extend trust transitively through them to people they've vetted, without personally verifying each one — that chain of person-to-person trust is the "web" in web of trust. It has a real, acknowledged weakness: it works best with a reasonably large, interconnected community of participants who actually do the vetting, and it breaks down when there aren't enough people using it for the chains to actually connect two strangers.^{TechTarget, Apr 2025} On darknet markets specifically, the web of trust rarely applies in its original form at all — a buyer verifying a market's key isn't building a web of personal vetting relationships, they're checking one specific key against one specific claimed identity, which is closer to the narrower fingerprint-comparison practice covered on PGP on the dark web than to the community trust model PGP was originally built around.

ASCII armor: why keys and messages look like that block of text

The actual output of PGP's cryptographic operations is raw binary data, which doesn't survive being copied and pasted through email clients, forum posts, and web forms reliably — some of those systems assume text and can corrupt raw binary silently. ASCII armor, also called Radix-64 encoding, converts that binary output into printable text characters bounded by a recognizable header and footer, specifically so a key or a signed message can be copied through any text-based channel without corruption.^{TechTarget, Apr 2025} The -----BEGIN PGP...----- and -----END PGP...----- lines shown above aren't decorative — they're exactly what lets software (and a human) reliably locate where the actual cryptographic content starts and ends inside a longer message or forum post.

Is PGP actually obsolete? What one widely-cited source gets wrong

One of the most-cited explainers for this exact query states plainly that "PGP has fallen in relevance"^{TechTarget, Apr 2025} — true in the narrow sense that mainstream consumer email encryption never took off the way its creators hoped, and false in the sense that matters for this site's readers: PGP is exactly as actively maintained and current as any cryptographic standard gets. The current OpenPGP specification, RFC 9580, was published in July 2024, superseding the 2007-era RFC 4880 that most older explainers, tutorials, and screenshots still describe.^Wikipedia The situation on the ground is genuinely more tangled than "one old standard, one new one," though: GnuPG's own longtime maintainer, Werner Koch, forked his implementation into a competing specification called LibrePGP rather than adopt RFC 9580, and announced GnuPG would not support the new standard going forward.^Wikipedia Practically, that means the single most widely used PGP tool on most people's machines is not implementing the newest official standard at all — it's running a related but distinct fork, which is a genuinely current, unresolved split in the ecosystem that a page calling PGP simply "less relevant" never mentions.